Scribd
Upload
3 views3 pages

Deep Learning Detailed Notes

Deep Learning is a subset of Machine Learning that utilizes Artificial Neural Networks to automatically learn patterns from large datasets. It differs from traditional Machine Learning by eliminating the need for feature engineering and employing more complex models. Key components include hyperparameters like epoch, batch size, learning rate, optimizer, activation function, and loss function, which are crucial for model training and accuracy.

Uploaded by

Sivasankaran Palanivel
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
3 views3 pages

Deep Learning Detailed Notes

Deep Learning is a subset of Machine Learning that utilizes Artificial Neural Networks to automatically learn patterns from large datasets. It differs from traditional Machine Learning by eliminating the need for feature engineering and employing more complex models. Key components include hyperparameters like epoch, batch size, learning rate, optimizer, activation function, and loss function, which are crucial for model training and accuracy.

Uploaded by

Sivasankaran Palanivel
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

Deep Learning - Complete Beginner Guide

1. What is Deep Learning?

Deep Learning is a subset of Machine Learning that uses Artificial Neural Networks with multiple
layers to learn patterns from data automatically.

Example: Predicting house prices using size, location, and rooms.

Real World: Face unlock in smartphones, voice assistants, self-driving cars.

2. Machine Learning vs Deep Learning

Machine Learning Deep Learning


Needs feature engineering Learns features automatically
Works with small data Needs large data
Less complex models Deep neural networks
Faster training Slower but powerful

3. What is Neural Network?

A Neural Network is a system inspired by the human brain that consists of layers of neurons.

It has three main layers: Input Layer, Hidden Layers, and Output Layer.

Example: Image → Model → Dog/Cat classification.

Real World: Google Photos identifying people and objects.

4. Working of Neural Network

Step 1: Input data is fed into the model.

Step 2: Forward propagation computes predictions.

Step 3: Loss function calculates error.

Step 4: Backpropagation updates weights.

Step 5: Process repeats until model improves.


Real World: Learning from mistakes like practicing exams repeatedly.

5. Important Hyperparameters

Epoch

Number of times the model sees the full dataset.

Example: 1000 rows, epoch=5 means model sees data 5 times.

Real World: Revising syllabus multiple times.

Batch Size

Number of samples processed at once.

Example: Batch size 100 means processing 100 rows at a time.

Real World: Studying chapter by chapter instead of full book.

Learning Rate

Controls how much weights change during learning.

Example: Small value = slow learning, large value = unstable.

Real World: Walking vs jumping to reach a destination.

Optimizer

Algorithm used to update weights to minimize loss.

Example: Adam, SGD.

Real World: Choosing best strategy to solve a problem efficiently.

Activation Function

Decides whether a neuron should activate or not.

Example: ReLU, Sigmoid.


Real World: Brain deciding whether to react to a signal.

Loss Function

Measures how wrong the model prediction is.

Example: Mean Squared Error, Cross Entropy.

Real World: Checking mistakes in an exam.

Final Summary

Deep Learning uses Neural Networks to learn patterns. Hyperparameters control how the model
learns, and tuning them properly helps achieve better accuracy.

You might also like